Proposes PARPO for reward decoupling with user anchors and PSGM for preference-aligned skill memory in personalized agentic RL, reporting outperformance on ETAPP benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
From Correctness to Preference: A Framework for Personalized Agentic Reinforcement Learning
Proposes PARPO for reward decoupling with user anchors and PSGM for preference-aligned skill memory in personalized agentic RL, reporting outperformance on ETAPP benchmarks.