Pith. sign in

Drift : Decoding-time personalized alignments with implicit user preferences

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

1 Pith paper citing it

fields

cs.CL 1

years

2026 1

verdicts

UNVERDICTED 1

representative citing papers

REAR: Test-time Preference Realignment through Reward Decomposition

cs.CL · 2026-06-29 · unverdicted · novelty 5.0

REAR decomposes the reward into question and preference components, rescales their balance, and expresses the result as a linear combination of token log-probabilities for efficient integration with best-of-N and tree search.

citing papers explorer

Showing 1 of 1 citing paper.

  • REAR: Test-time Preference Realignment through Reward Decomposition cs.CL · 2026-06-29 · unverdicted · none · ref 117

    REAR decomposes the reward into question and preference components, rescales their balance, and expresses the result as a linear combination of token log-probabilities for efficient integration with best-of-N and tree search.