A collection of BERT-based models of different sequence lengths trained on the C4 dataset. Details: https://mosaicbert.github.io/