alphaXiv @askalphaxiv · 02 Sep 2026
“Knowledge Distillation During Mid-Training Favors Reasoning over Factual Recall” Knowledge distillation helps reasoning during mid-training, but surprisingly hurts factual recall because teachers are much less confident on the remaining knowledge-heavy tokens. So this paper https://t.co/IvWO0963Dc
3 426Views
57Likes
5Reposts
1Replies
0Quotes
46Bookmarks
Is that a lot?
1.28×vs this author's median2 682 views is typical
50Percentile for this authorof 4 recent posts
1.18×vs 10K–100K median2 896 views is typical
6.35%Reachviews ÷ followers
1.84%Engagement rateof viewers reacted