HEAL mitigates entropy collapse in few-shot RLVR by selectively adding general-domain data and aligning trajectory-level entropy dynamics, matching full-shot performance with 32 target samples.
Active preference learning for large language models
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2verdicts
UNVERDICTED 2representative citing papers
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
-
HEALing Entropy Collapse: Enhancing Exploration in Few-Shot RLVR via Hybrid-Domain Entropy Dynamics Alignment
HEAL mitigates entropy collapse in few-shot RLVR by selectively adding general-domain data and aligning trajectory-level entropy dynamics, matching full-shot performance with 32 target samples.
-
Test-Time Alignment via Hypothesis Reweighting
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.