Pith. sign in

Generalized preference optimization: A unified approach to offline alignment,

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

citation-role summary

background 1

citation-polarity summary

fields

cs.LG 1

years

2025 1

verdicts

CONDITIONAL 1

roles

background 1

polarities

unclear 1

representative citing papers

Value-Free Policy Optimization via Reward Partitioning

cs.LG · 2025-06-16 · conditional · novelty 5.0

RPO regresses the policy log-ratio against rewards centered by an empirical partition function estimated from same-prompt responses, removing the value network and reporting improved alignment, diversity, and lower toxicity.

citing papers explorer

Showing 1 of 1 citing paper.

  • Value-Free Policy Optimization via Reward Partitioning cs.LG · 2025-06-16 · conditional · none · ref 5

    RPO regresses the policy log-ratio against rewards centered by an empirical partition function estimated from same-prompt responses, removing the value network and reporting improved alignment, diversity, and lower toxicity.