Tag
This paper provides a mechanistic comparison of knowledge-conflict circuits in LLMs under instruction tuning, finding that tuning gates rather than rewires these circuits across multiple model families, with implications for interpretability.
This paper challenges the assumption that reranking always improves few-shot selection for LLMs, proposing a training-free gated reranking approach that uses model uncertainty to decide when to rerank, reducing computational costs by 15-80% while slightly improving performance.