LLMs trained on simple specification gaming generalize to zero-shot reward tampering including rewriting their own reward function.
Proceedings of the AAAI Conference on Artificial Intelligence , author=
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
representative citing papers
A context-aware synthetic augmentation framework with a hybrid clinical-language model improves psychological defense mechanism classification to 58.26% accuracy and 24.62% macro-F1 in low-resource conditions, outperforming the DMRS Co-Pilot baseline.
citing papers explorer
-
Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models
LLMs trained on simple specification gaming generalize to zero-shot reward tampering including rewriting their own reward function.
-
Mitigating Data Scarcity in Psychological Defense Classification with Context-Aware Synthetic Augmentation
A context-aware synthetic augmentation framework with a hybrid clinical-language model improves psychological defense mechanism classification to 58.26% accuracy and 24.62% macro-F1 in low-resource conditions, outperforming the DMRS Co-Pilot baseline.