Superposition

TagLast edit: Dec 5, 2023, 8:41 PM by duck_master

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

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

Lee Sharkey, Dan Braun and beren

Dec 13, 2022, 3:41 PM

150 points

23 comments22 min readLW link 2 reviews

Superposition is not “just” neuron polysemanticity

LawrenceCApr 26, 2024, 11:22 PM

66 points

4 comments13 min readLW link

Toward A Mathematical Framework for Computation in Superposition

Dmitry Vaintrob, jake_mendel and Kaarel

Jan 18, 2024, 9:06 PM

205 points

18 comments63 min readLW link

Circuits in Superposition: Compressing many small neural networks into one

Lucius Bushnaq and jake_mendel

Oct 14, 2024, 1:06 PM

130 points

9 comments13 min readLW link

Conditional Importance in Toy Models of Superposition

james__pFeb 2, 2025, 8:35 PM

9 points

4 comments10 min readLW link

Some costs of superposition

Linda LinseforsMar 3, 2024, 4:08 PM

46 points

11 comments3 min readLW link

Thoughts on Toy Models of Superposition

james__pFeb 2, 2025, 1:52 PM

5 points

2 comments9 min readLW link

AI alignment as a translation problem

Roman LeventovFeb 5, 2024, 2:14 PM

22 points

2 comments3 min readLW link

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

Roman LeventovFeb 6, 2024, 10:18 AM

8 points

1 comment4 min readLW link

(arxiv.org)

An OV-Coherent Toy Model of Attention Head Superposition

Lauren Greenspan and keith_wynroe

Aug 29, 2023, 7:44 PM

26 points

2 comments6 min readLW link

Sparse autoencoders find composed features in small toy models

Evan Anders, Clement Neo, Jason Hoelscher-Obermaier and Jessica N. Howard

Mar 14, 2024, 6:00 PM

33 points

12 comments15 min readLW link

Superposition through Active Learning Lens

akankshancSep 17, 2024, 5:32 PM

1 point

0 comments10 min readLW link

Effects of Non-Uniform Sparsity on Superposition in Toy Models

Shreyans JainNov 14, 2024, 4:59 PM

4 points

3 comments6 min readLW link

Scaling Laws and Superposition

Pavan KattaApr 10, 2024, 3:36 PM

9 points

4 comments5 min readLW link

(www.pavankatta.com)

Expanding the Scope of Superposition

Derek LarsonSep 13, 2023, 5:38 PM

10 points

0 comments4 min readLW link

Some open-source dictionaries and dictionary learning infrastructure

Sam MarksDec 5, 2023, 6:05 AM

46 points

7 comments5 min readLW link

Toy Models of Superposition

evhubSep 21, 2022, 11:48 PM

69 points

4 comments5 min readLW link 1 review

(transformer-circuits.pub)

Superposition and Dropout

Edoardo PonaMay 16, 2023, 7:24 AM

21 points

5 comments6 min readLW link

Growth and Form in a Toy Model of Superposition

Liam Carroll and Edmund Lau

Nov 8, 2023, 11:08 AM

89 points

7 comments14 min readLW link

Paper: Superposition, Memorization, and Double Descent (Anthropic)

LawrenceCJan 5, 2023, 5:54 PM

53 points

11 comments1 min readLW link

(transformer-circuits.pub)

200 COP in MI: Exploring Polysemanticity and Superposition

Neel NandaJan 3, 2023, 1:52 AM

34 points

6 comments16 min readLW link

Towards Monosemanticity: Decomposing Language Models With Dictionary Learning

Zac Hatfield-DoddsOct 5, 2023, 9:01 PM

288 points

22 comments2 min readLW link 1 review

(transformer-circuits.pub)

Limitations on the Interpretability of Learned Features from Sparse Dictionary Learning

Tom AngstenJul 30, 2024, 4:36 PM

6 points

0 comments9 min readLW link

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

Joseph BloomFeb 2, 2024, 6:54 AM

103 points

37 comments15 min readLW link

Comparing Anthropic’s Dictionary Learning to Ours

Robert_AIZIOct 7, 2023, 11:30 PM

137 points

8 comments4 min readLW link

Computational Superposition in a Toy Model of the U-AND Problem

Adam NewgasMar 27, 2025, 4:56 PM

17 points

2 comments11 min readLW link

Toy Models of Superposition: Simplified by Hand

Axel SorensenSep 29, 2024, 9:19 PM

9 points

3 comments8 min readLW link

Taking features out of superposition with sparse autoencoders more quickly with informed initialization

Pierre PeignéSep 23, 2023, 4:21 PM

30 points

8 comments5 min readLW link

Sparse MLP Distillation

slavachalnevJan 15, 2024, 7:39 PM

30 points

3 comments6 min readLW link

Crafting Polysemantic Transformer Benchmarks with Known Circuits

Evan Anders and Adrià Garriga-alonso

Aug 23, 2024, 10:03 PM

17 points

0 comments25 min readLW link

Interpretability with Sparse Autoencoders (Colab exercises)

CallumMcDougallNov 29, 2023, 12:56 PM

76 points

9 comments4 min readLW link

No comments.