© 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 · 2695 ｜ Neuronpedia

Home
GPT2-Small
Transcoders Residuals
8-TRES-DC
2695

INDEX

Explanations

punctuation marks and their implications in context

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

foundland

-0.79

lectic

-0.71

CLOSE

-0.70

oday

-0.68

leneck

-0.68

cedented

-0.67

ometown

-0.66

nown

-0.65

rolet

-0.64

ews

-0.64

POSITIVE LOGITS

it

0.76

why

0.68

 Hats

0.66

 critics

0.63

 myths

0.63

 caveats

0.63

 virtues

0.62

 campaigners

0.61

 observers

0.61

 results

0.60

Activations Density 0.154%

No Known Activations