Papers with Code - Cross-lingual Language Model Pretraining
š SOTA for Unsupervised Machine Translation on WMT2016 English--Romanian (BLEU metric)
facebookresearch/XLM. 2,300. facebookresearch/ ... Get a GitHub badge ... Natural Language Inference, XNLI French, XLM (MLM+TLM), Accuracy, 80.2, # 2.