2022

Polysemanticity and Capacity in Neural Networks

Scherlis, Adam, Sachan, Kshitij, Jermyn, Adam S. et al.

Understand

Individual neurons in neural networks often represent a mixture of unrelated features.

  • This phenomenon, called polysemanticity, can make interpreting neural networks more difficult and so we aim to understand its causes.
  • We propose doing so through the lens of feature \emph{capacity}, which is the fractional dimension each feature consumes in the embedding space.
  • We show that in a toy model the optimal capacity allocation tends to monosemantically represent the most important features, polysemantically represent less important features (in proportion to their impact on the loss), and entirely ignore the least important features.

Built on

  • Feature visualization

    Chris Olah, Alexander Mordvintsev, and Ludwig Schubert · 2017

    Earlier work this paper cites.

  • Zoom in: An introduction to circuits

    Chris Olah, Nick Cammarata, Ludwig Schubert, Gabriel Goh, Michael Petrov, and Shan Carter · 2020

    Earlier work this paper cites.

Similar

  • Softmax linear units

    Nelson Elhage, Tristan Hume, Catherine Olsson, Neel Nanda, Tom Henighan, Scott Johnston, Sheer ElShowk, Nicholas Joseph, Nova DasSarma, Ben Mann, Danny Hernandez, Amanda Askell, Kamal Ndousse, Andy Jones, Dawn Drain, Anna Chen, Yuntao Bai, Deep Ganguli, Liane Lovitt, Zac Hatfield-Dodds, Jackson Kernion, Tom Conerly, Shauna Kravec, Stanislav Fort, Saurav Kadavath, Josh Jacobson, Eli Tran-Johnson, Jared Kaplan, Jack Clark, Tom Brown, Sam McCandlish, Dario Amodei, and Christopher Olah · 2022

    Cited alongside, same era.

Then

  • Toy models of superposition

    Nelson Elhage, Tristan Hume, Catherine Olsson, Nicholas Schiefer, Tom Henighan, Shauna Kravec, Zac Hatfield-Dodds, Robert Lasenby, Dawn Drain, Carol Chen, Roger Grosse, Sam McCandlish, Jared Kaplan, Dario Amodei, Martin Wattenberg, and Christopher Olah · 2022

    Closest in time.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…