Superposition in LLM Feature Representations

Superposition in LLM Feature Representations

Conf42 via YouTube Direct link

intro

1 of 18

1 of 18

intro

Class Central Classrooms beta

YouTube videos curated by Class Central.

Classroom Contents

Superposition in LLM Feature Representations

Automatically move to the next video in the Classroom when playback concludes

  1. 1 intro
  2. 2 preamble
  3. 3 mechanistic interpretability
  4. 4 neural network representations
  5. 5 qualities of representations
  6. 6 decomposability
  7. 7 linearity
  8. 8 linear composition as a compression scheme
  9. 9 demands of linearity
  10. 10 the linear representation puzzle
  11. 11 neuron - feature requirements
  12. 12 experience with llms
  13. 13 the superposition hypothesis
  14. 14 sparsity
  15. 15 recovering features in superposition
  16. 16 demands of linearity
  17. 17 feature exploration
  18. 18 thanks

Never Stop Learning.

Get personalized course recommendations, track subjects and courses with reminders, and more.

Someone learning on their laptop while sitting on the floor.