Combined views
144
2 posts, first seen 4h ago
15 reposts
144
2 posts, first seen 4h ago
Users are praising the Sparse Delta Memory research and its author's PhD work as amazing and great because it enables 3000x larger RNN states at unchanged FLOPs.
Based on 2 visible X reactions from 6 accounts; directional sample.
Ask a question below.
Published answers will appear here.
The technique matches KV cache recall using product key sparsity
Users are praising the Sparse Delta Memory research and its author's PhD work as amazing and great because it enables 3000x larger RNN states at unchanged FLOPs.
Based on 2 visible X reactions from 6 accounts; directional sample.
Ask a question below.
Published answers will appear here.