PaperLens
紙
Students
Professional
JA
EN
◐
Sign in with Google
Sign in
Read
Home
Close reading
New
Textbook
Go deeper
Learn
Lab
Landscape
Contributors
Glossary
You
Search
All-access
My Page
#test-time-training
1 articles
01
2026-08-31
·
Inference & Serving
·
★ MEMBER
·
PAPER
·
11 min read
TTPO Explained: Training a Model Mid-Exam, With No Answer Key
Majority-vote pseudo-labels are wrong on roughly 85% of competition math prompts. So why does training still work? A ground-up walkthrough of TTPO's asymmetric design: distill the rollouts that agree, penalize the ones that disagree.