I pretrained the best small model under 100M parameters

(twitter.com) by cybernovas | Jul 31, 2026 | 1 comments on HN