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
9-TRES-DC
620

INDEX

Explanations

references to the Metro system

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

othy

-0.77

 posing

-0.70

 thirds

-0.67

ento

-0.65

uran

-0.64

 ming

-0.63

auntlet

-0.62

oe

-0.61

 risked

-0.61

 trailing

-0.61

POSITIVE LOGITS

魔

0.66

WI

0.65

ibi

0.63

essee

0.63

Indian

0.63

�

0.62

Leader

0.62

ioned

0.62

nik

0.60

シャ

0.60

Activations Density 0.019%

No Known Activations