TY  - RPRT
TI  - PivotOPD: Learning to Recover from Pivotal Mistakes in Multi-Turn Agents
AU  - Yinghui He
AU  - Yapei Chang
AU  - Khushi Bhardwaj
AU  - Daniele Molinari
AU  - Tugrul Konuk
AU  - Jan Kautz
AU  - Ali Hatamizadeh
PY  - 2026
UR  - https://arxiv.org/abs/2609.40285
ID  - 2609.40285
ER  -