Superposition

Edited by duck_master last updated 5th Dec 2023

Posts about the concept of superposition - that is, neural nets representing concepts as a superposition of many neurons.

Posts tagged Superposition

8

155[Interim research report] Taking features out of superposition with sparse autoencoders

Ω

Lee Sharkey, Dan Braun, beren

3y

Ω

23

2

212Toward A Mathematical Framework for Computation in Superposition

Ω

Dmitry Vaintrob, jake_mendel, Kaarel

2y

Ω

19

2

131Circuits in Superposition: Compressing many small neural networks into one

Ω

Lucius Bushnaq, jake_mendel

1y

Ω

9

2

73Circuits in Superposition 2: Now with Less Wrong Math

Ω

Linda Linsefors, Lucius Bushnaq

5mo

Ω

0

2

69Superposition is not "just" neuron polysemanticity

Ω

LawrenceC

2y

Ω

4

2

46Some costs of superposition

Ω

Linda Linsefors

2y

Ω

11

2

23AI alignment as a translation problem

Roman Leventov

2y

2

9Conditional Importance in Toy Models of Superposition

james__p

10mo

4

2

8From Conceptual Spaces to Quantum Concepts: Formalising and Learning Structured Conceptual Models

Roman Leventov

2y

1

2

5Thoughts on Toy Models of Superposition

james__p

10mo

2

1

289Towards Monosemanticity: Decomposing Language Models With Dictionary Learning

Ω

Zac Hatfield-Dodds

2y

Ω

22

1

137Comparing Anthropic's Dictionary Learning to Ours

Ω

Robert_AIZI

2y

Ω

8

1

103Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small

Ω

Joseph Bloom

2y

Ω

37

1

90Growth and Form in a Toy Model of Superposition

Ω

Liam Carroll, Edmund Lau

2y

Ω

7

1

79Interpretability with Sparse Autoencoders (Colab exercises)

Ω

CallumMcDougall

2y

Ω

9

LESSWRONG
LW

LESSWRONG
LW

Superposition