HyperAI초신경
홈뉴스최신 연구 논문튜토리얼데이터셋백과사전SOTALLM 모델GPU 랭킹컨퍼런스
전체 검색
소개
HyperAI초신경

Machine Translation On Wmt2014 German English

평가 지표

BLEU score

평가 결과

이 벤치마크에서 각 모델의 성능 결과

모델 이름
BLEU score
Paper TitleRepository
CMLM+LAT+1 iterations29.91Incorporating a Local Translation Mechanism into Non-autoregressive Translation-
SimCut34.86Bi-SimCut: A Simple Strategy for Boosting Neural Machine Translation
CNAT30.75Non-Autoregressive Translation by Learning Target Categorical Codes
BiBERT34.94BERT, mBERT, or BiBERT? A Study on Contextualized Embeddings for Neural Machine Translation
Denoising autoencoders (non-autoregressive)25.43Deterministic Non-Autoregressive Neural Sequence Modeling by Iterative Refinement
NAT +FT + NPD23.20Non-Autoregressive Neural Machine Translation
FlowSeq-large (NPD n = 15)27.71FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow
Mega33.12Mega: Moving Average Equipped Gated Attention
SMT + iterative backtranslation (unsupervised)17.43Unsupervised Statistical Machine Translation
FlowSeq-large25.4FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow
Bi-SimCut35.15Bi-SimCut: A Simple Strategy for Boosting Neural Machine Translation
FlowSeq-base23.36FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow
MAT+Knee31.9Wide-minima Density Hypothesis and the Explore-Exploit Learning Rate Schedule
CMLM+LAT+4 iterations32.04Incorporating a Local Translation Mechanism into Non-autoregressive Translation-
FlowSeq-large (IWD n=15)27.16FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow
FlowSeq-large (NPD n = 30)28.29FlowSeq: Non-Autoregressive Conditional Sequence Generation with Generative Flow
0 of 16 row(s) selected.