Causal reward modeling applies MMD regularization to RLHF reward models to make reward scores statistically independent of spurious features, reducing measured length, sycophancy, concept, and demographic biases in experiments.
Selection Bias Induced Spurious Correlations in Large Language Models
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
In this work we show how large language models (LLMs) can learn statistical dependencies between otherwise unconditionally independent variables due to dataset selection bias. To demonstrate the effect, we developed a masked gender task that can be applied to BERT-family models to reveal spurious correlations between predicted gender pronouns and a variety of seemingly gender-neutral variables like date and location, on pre-trained (unmodified) BERT and RoBERTa large models. Finally, we provide an online demo, inviting readers to experiment further.
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
Beyond Reward Hacking: Causal Rewards for Large Language Model Alignment
Causal reward modeling applies MMD regularization to RLHF reward models to make reward scores statistically independent of spurious features, reducing measured length, sycophancy, concept, and demographic biases in experiments.