Crome trains reward models on LLM-generated causal and neutral augmentations, improving RewardBench accuracy by up to 5.4% and robustness to spurious transformations.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Robust Reward Modeling via Causal Rubrics
Crome trains reward models on LLM-generated causal and neutral augmentations, improving RewardBench accuracy by up to 5.4% and robustness to spurious transformations.