Seohong Park @seohong_park · 24 Aug 2026
@seohong_park Good paper on why action chunking works: https://t.co/B7dQIyoBGd It seems to me it's because human generated trajectories are non-Markovian. Maybe because humans plan and execute at different frequencies. But curious if the same phenomenon occurs on synthetic data.
1 342Views
10Likes
0Reposts
0Replies
0Quotes
9Bookmarks
Is that a lot?
0.89×vs this author's median1 510 views is typical
25Percentile for this authorof 4 recent posts
0.50×vs under 10K median2 700 views is typical
23.21%Reachviews ÷ followers
0.74%Engagement rateof viewers reacted