conftrace_

sparse activation editing

1 paper

Explore in graph

Also known as

SAE

Co-occurring keywords · papers mentioning both

neuron editing (1) activation editing (1) training-free method (1) mechanistic interpretability (1) sparse activation (1) language model (1) model editing (1) narrative understanding (1) instruction following (1)

Papers

Sparse Activation Editing for Reliable Instruction Following in Narratives EMNLP 2025