“The Local Interaction Basis: Identifying Computationally-Relevant and Sparsely Interacting Features in Neural Networks” by Marius Hobbhahn, Lucius Bushnaq, jake_mendel, Dan Braun, StefanHex, Nicholas Goldowsky-Dill, Kaarel, Avery, Joern Stoehler, debrevitatevitae, Magdalena Wache

LessWrong (30+ Karma) - “The Local Interaction Basis: Identifying Computationally-Relevant and Sparsely Interacting Features in Neural Networks” by Marius Hobbhahn, Lucius Bushnaq, jake_mendel, Dan Braun, StefanHex, Nicholas Goldowsky-Dill, Kaarel, Avery, Joern Stoehler, debrevitatevitae, Magdalena Wache