elie @eliebakouch · 02 Sep 2026
this is not a "bitter lesson" pilled way to scale transformers, you're introducing a structural bias by looping blocks instead of having more unique weights and letting them specialize as they want
16 476Views
189Likes
6Reposts
16Replies
0Quotes
76Bookmarks
Is that a lot?
0.52×vs this author's median31 519 views is typical
29Percentile for this authorof 7 recent posts
5.99×vs 10K–100K median2 751 views is typical
68.33%Reachviews ÷ followers
1.28%Engagement rateof viewers reacted