pith. sign in

Regularizing hidden states enables learning generalizable reward model for llms

2 Pith papers cite this work. Polarity classification is still indexing.

2 Pith papers citing it

fields

cs.AI 1 cs.LG 1

years

2024 2

verdicts

UNVERDICTED 2

representative citing papers

Test-Time Alignment via Hypothesis Reweighting

cs.LG · 2024-12-11 · unverdicted · novelty 5.0

HyRe personalizes reward models at test time by reweighting an ensemble of heads trained on aggregate preferences, using few target examples to outperform uniform averaging and prior methods on RewardBench and 32 tasks.

citing papers explorer

Showing 2 of 2 citing papers.

  • Test-Time Alignment via Hypothesis Reweighting cs.LG · 2024-12-11 · unverdicted · none · ref 64

    HyRe personalizes reward models at test time by reweighting an ensemble of heads trained on aggregate preferences, using few target examples to outperform uniform averaging and prior methods on RewardBench and 32 tasks.

  • Skywork-Reward: Bag of Tricks for Reward Modeling in LLMs cs.AI · 2024-10-24 · unverdicted · none · ref 24

    Data-centric filtering yields an 80K preference dataset and reward models that lead RewardBench while boosting other top entries.