H3 – Outperforming GPT-Neo-2.7B with only 2 attention layers (twitter.com) 10 points by m00x 3y ago ↗ HN
0 comments
[ 4.0 ms ] story [ 10.6 ms ] threadNo comments yet.