[NeurIPS 2021] COCO-LM: Correcting and Contrasting Text Sequences for Language Model Pretraining
We release two pretrained COCO-LM model checkpoints and one dictionary file:
cocolm-base.tar.gz contains the COCO-LM base++ model; you need to extract the model from the archive.cocolm-large.tar.gz contains the COCO-LM large++ model; you need to extract the model from the archive.dict.tar.gz contains the sentencepiece model (sp.model) and the vocabulary file (dict.txt).