Par-S²ZPO matches centralized RLHF sample complexity while converging faster in policy updates and outperforming FedAvg on MuJoCo tasks.
signrx∇ θVpπ θtq, µtvt,kys “1 . The same argument applies to where x∇θVpπ θtq, µtvt,ky ă0. As a consequence, we have onE c t,k that: A ∇θV pπθtq,sign
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Efficient Federated RLHF via Zeroth-Order Policy Optimization
Par-S²ZPO matches centralized RLHF sample complexity while converging faster in policy updates and outperforming FedAvg on MuJoCo tasks.