© 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 · 2893 ｜ Neuronpedia

Home
GPT2-Small
Transcoders Residuals
8-TRES-DC
2893

INDEX

Explanations

variations of the word "darn" and related expressions

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

DragonMagazine

-0.48

anamo

-0.46

agram

-0.40

CVE

-0.38

fman

-0.37

gov

-0.37

thouse

-0.36

oeuv

-0.36

Lago

-0.36

stanbul

-0.34

POSITIVE LOGITS

Beg

0.38

Des

0.35

 Proc

0.33

Ath

0.33

 Commit

0.33

ap

0.32

 Param

0.32

Ask

0.32

Pse

0.32

 Inner

0.32

Activations Density 7.909%

No Known Activations