© 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 · 2649 ｜ Neuronpedia

Home
GPT2-Small
Transcoders Residuals
8-TRES-DC
2649

INDEX

Explanations

phrases indicating future timing or anticipation

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

IONS

-0.64

clud

-0.61

CBI

-0.60

irin

-0.60

 Others

-0.58

 absent

-0.55

Hok

-0.55

aths

-0.55

okers

-0.55

 sexes

-0.54

POSITIVE LOGITS

 soon

0.86

late

0.75

uyomi

0.74

soon

0.69

olulu

0.69

olid

0.64

sequently

0.64

Soon

0.64

uture

0.63

 late

0.63

Activations Density 0.033%

No Known Activations