© Neuronpedia 2026

Privacy & Terms Blog GitHub Slack Twitter Contact

Neuronpedia

Jacobian LensNEW

Natural Language

NEW Assistant AxisNEW Circuit TracerUPDATESteer SAE Evals Exports interp-engineNEWAPI Community Blog Privacy & Terms Contact

GPT2-SMALL · 8-TRES-DC · 2928 ｜ Neuronpedia

Home
GPT2-Small
Transcoders Residuals
8-TRES-DC
2928

INDEX

Explanations

references to software updates and feature enhancements

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

New Auto-Interp

Top Features by Cosine Similarity

Embeds

Show PlotsShow ExplanationShow ActivationsShow Test FieldShow SteerShow Link

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 wonder

-0.72

 cheers

-0.66

 commuting

-0.65

 directions

-0.63

 propri

-0.60

 refuel

-0.60

 instructions

-0.59

MIL

-0.59

 tweets

-0.59

 resort

-0.59

POSITIVE LOGITS

rang

0.73

uilt

0.73

acus

0.72

lav

0.72

icable

0.69

installed

0.69

vim

0.68

agically

0.68

emaker

0.68

este

0.67

Activations Density 0.400%

No Known Activations