Machine Translation On Wmt2014 English French

评估指标

BLEU score

评测结果

各个模型在此基准测试上的表现结果

		Paper Title
Transformer+BT (ADMIN init)	46.4	Very Deep Transformers for Neural Machine Translation
Noisy back-translation	45.6	Understanding Back-Translation at Scale
mRASP+Fine-Tune	44.3	Pre-training Multilingual Neural Machine Translation by Leveraging Alignment Information
Transformer + R-Drop	43.95	R-Drop: Regularized Dropout for Neural Networks
Admin	43.8	Understanding the Difficulty of Training Transformers
Transformer (ADMIN init)	43.8	Very Deep Transformers for Neural Machine Translation
BERT-fused NMT	43.78	Incorporating BERT into Neural Machine Translation
MUSE(Paralllel Multi-scale Attention)	43.5	MUSE: Parallel Multi-Scale Attention for Sequence to Sequence Learning
T5	43.4	Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
Local Joint Self-attention	43.3	Joint Source-Target Self Attention with Locality Constraints
Depth Growing	43.27	Depth Growing for Neural Machine Translation
DynamicConv	43.2	Pay Less Attention with Lightweight and Dynamic Convolutions
Transformer Big	43.2	Scaling Neural Machine Translation
TaLK Convolutions	43.2	Time-aware Large Kernel Convolutions
LightConv	43.1	Pay Less Attention with Lightweight and Dynamic Convolutions
FLOATER-large	42.7	Learning to Encode Position for Transformer with Continuous Dynamical Model
OmniNetP	42.6	OmniNet: Omnidirectional Representations from Transformers
T2R + Pretrain	42.1	Finetuning Pretrained Transformers into RNNs
Transformer Big + MoS	42.1	Fast and Simple Mixture of Softmaxes with BPE and Hybrid-LightRNN for Language Generation
Synthesizer (Random + Vanilla)	41.85	Synthesizer: Rethinking Self-Attention in Transformer Models

0 of 57 row(s) selected.