© Neuronpedia 2026

Privacy & Terms Blog GitHub Slack Twitter Contact

Neuronpedia

Natural Language

NEW Assistant AxisNEW Circuit TracerUPDATESteer SAE Evals ExportsAPI Community Blog Privacy & Terms Contact

GPT2-SMALL · 8-TRES-DC · 1395 ｜ Neuronpedia

Home
GPT2-Small
Transcoders Residuals
8-TRES-DC
1395

INDEX

Explanations

phrases related to consecutive sequences or streaks

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

undo

-0.83

avior

-0.80

CHAT

-0.79

roma

-0.78

SAY

-0.75

SELECT

-0.74

OLOGY

-0.74

VD

-0.74

ewitness

-0.71

galitarian

-0.71

POSITIVE LOGITS

 nights

0.97

 evenings

0.87

 mornings

0.86

 winters

0.83

 Sundays

0.80

 summers

0.79

 Tornado

0.75

 weekends

0.71

 weeks

0.71

 Kelley

0.69

Activations Density 0.053%

No Known Activations