TY  - RPRT
TI  - Trust Functions: Near-Lossless Weak-to-Strong Generalization by Learning When to Trust the Weak Teacher
AU  - Arda Uzunoglu
AU  - Alvin Zhang
AU  - Daniel Khashabi
PY  - 2026
UR  - https://arxiv.org/abs/2606.01000
ID  - 2606.01000
ER  -