lucidrains / memorizing-transformers-pytorch

Implementation of Memorizing Transformers (ICLR 2022), attention net augmented with indexing and retrieval of memories using approximate nearest neighbors, in Pytorch
MIT License
625 stars 46 forks source link