Large Open-Weight MoEs Are Converging on Lower Expert Activation Ratios (twitter.com)

1 points by ayushpai 3mo ago ↗ HN

0 comments

[ 2.8 ms ] story [ 9.5 ms ] thread

No comments yet.

A showcase of shard-db. Data from the Hacker News API. Refreshed every 15 minutes.

Source · Live DB stats