AI Alignment Forum company blog
A Mechanistic Interpretability Analysis of Grokking
Neel Nanda reverse engineers neural networks that have "grokked" modular addition, showing that they operate using Discrete Fourier Transforms and tr…
- Points
- 202
- HN comments
- 54
- Submissions
- 1
- Cited by
- 5 commenters
AI & Machine LearningMath & Algorithms2023
On Hacker News
| Submitted | Title on HN | Points | Comments |
|---|---|---|---|
| May 30, 2023 | A Mechanistic Interpretability Analysis of Grokking by famouswaffles | 202 | 54 |
Cited in comments
Linked from 11 Hacker News comments by 5 different people, from Jan 2023 to Oct 2024.
Most recent:
- famouswaffles · Oct 27, 2024
- famouswaffles · May 29, 2024
- famouswaffles · May 25, 2024
- og_kalu · Mar 25, 2024
- og_kalu · Feb 28, 2024
- og_kalu · Jan 8, 2024
