Signal65 @Signal_65 · 03 Sep 2026
@Signal_65 Interesting that 3.5x the output tokens nearly catches up in quality. Probably a big reason for the slowdown in the decode phase, but Contemplating seems worth it!
42Views
1Likes
0Reposts
0Replies
0Quotes
0Bookmarks
Is that a lot?
0Percentile for this authorof 3 recent posts
0.01×vs under 10K median3 808 views is typical
3.78%Reachviews ÷ followers
2.38%Engagement rateof viewers reacted