Neuronpedia

APISteer SAE Evals Blog/PodcastNEW Slack Privacy & Terms Contact

© Neuronpedia 2025

Privacy & Terms Blog/Podcast GitHub Slack Twitter Contact

Home
Dunefsky · Chlenski · Transcoders Enable Fine-Grained Interpretable Circuit Analysis
GPT2-Small
Transcoders Residuals
8-TRES-DC
18

INDEX

Explanations

geographical locations or names

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

CLASSIFIED

-0.70

 Almighty

-0.67

>[

-0.66

Touch

-0.64

TVs

-0.63

Volume

-0.63

 Tablet

-0.62

CLR

-0.60

 Univ

-0.60

 Volume

-0.60

POSITIVE LOGITS

ossal

0.88

folk

0.88

ulic

0.87

alian

0.81

wed

0.81

olen

0.78

angan

0.78

uters

0.78

chin

0.78

omp

0.76

Activations Density 0.117%

No Known Activations