Ivan ChelombievinTowards Data ScienceGetting more bang for your buck out of pre-trained language models with GroupBERTAchieving 2x training speed-up on the Graphcore IPUJan 21, 2022Jan 21, 2022