Jason Wei @_jasonwei · 16 Jul 2025
@_jasonwei We actually have a justification for this very intuition "I would never surpass the full ability of someone else because they were playing to their strengths which I didn’t have" -- we show that RL can adapt/calibrate CoTs to the model's own capability, while SFT often leads to
14 170Views
64Likes
8Reposts
1Replies
0Quotes
26Bookmarks
Is that a lot?
0.31×vs this author's median45 457 views is typical
38Percentile for this authorof 8 recent posts
3.44×vs 100K–1M median4 116 views is typical
12.56%Reachviews ÷ followers
0.52%Engagement rateof viewers reacted