TY  - RPRT
TI  - An Imperfect Verifier is Good Enough: Learning with Noisy Rewards
AU  - Andreas Plesner
AU  - Francisco Guzmán
AU  - Anish Athalye
PY  - 2026
UR  - https://arxiv.org/abs/2604.07666
ID  - 2604.07666
ER  -