Google DeepMind
@googledeepmind
We’re a team of scientists, engineers, ethicists and more, committed to solving intelligence, to advance science and benefit humanity.
ID: 4783690002
http://deepmind.google 19-01-2016 13:46:08
3,3K Tweet
1,1M Followers
276 Following
RFA is a linear-complexity attention based on random feature methods. Applied in transformer models, it improves efficiency w/o accuracy loss on MT and language modelling: dpmd.ai/RFA Work by @haopeng01, Nikos Pappas, Dani Yogatama, Roy Schwartz, Noah A. Smith & Lingpeng Kong
