Neuronpedia

Creator

Creator

Created

Created

2024 Sep 9 22:13

Editor

Editor

Edited

Edited

2025 Aug 5 23:55

Refs

Refs

neuronpedia
hijohnnylin • Updated 2025 Apr 6 17:50

automated-interpretability
hijohnnylin • Updated 2025 Jun 17 9:24

neuronpedia-python
hijohnnylin • Updated 2025 Jul 1 0:35

neuronpedia-scorer
hijohnnylin • Updated 2025 Apr 6 17:52

notion image

Automated Interpretability

SAE Steering

With a single click on Vercel, individuals and organizations can easily self-host, and freely modify functionality and UI by forking as desired

Description in readonly open DB

Activation in S3 parquet

`postgresql://readonly_user:np-public@34.123.166.159/postgres`

Model

Neuron

Explanation

Activation

Neuronpedia Usages

Neuronpedia Circuit Tracing

Open Interpretability Platform

https://www.neuronpedia.org/

Google DeepMind has a new way to look inside an AI’s “mind”

Autoencoders are letting us peer into the black box of artificial intelligence. They could help us create AI that is better understood, and more easily controlled.

https://www.technologyreview.com/2024/11/14/1106871/google-deepmind-has-a-new-way-to-look-inside-an-ais-mind/

Google DeepMind has a new way to look inside an AI’s “mind”

Activation

neuronpedia-exports.s3.amazonaws.com

https://neuronpedia-exports.s3.amazonaws.com/

init

Announcing Neuronpedia: Platform for accelerating research into Sparse Autoencoders — AI Alignment Forum

This posts assumes basic familiarity with Sparse Autoencoders. For those unfamiliar with this technique, we highly recommend the introductory section…

Announcing Neuronpedia: Platform for accelerating research into Sparse Autoencoders — AI Alignment Forum

https://www.alignmentforum.org/posts/BaEQoxHhWPrkinmxd/announcing-neuronpedia-platform-for-accelerating-research

Announcing Neuronpedia: Platform for accelerating research into Sparse Autoencoders — AI Alignment Forum

Backlinks

Circuit Tracing Mechanistic interpretability

Recommendations

//////////