News for hackers. Stuff that matters.
Re:Memorizing Transformers by silencedogood3 on Friday May 20, 2022 @12:34PM (#31449673)
Neat! Can you explain what the KNn is doing? I can’t quite follow the paper.
Re:Memorizing Transformers by visarga on Friday May 20, 2022 @12:49PM (#31449895)
It's a sparse attention scheme. They store and reuse activations thus "memorising" the past without the need for training. In order to keep the sequence short enough to fit into memory they only recall the k most similar memories from a much larger log.