© Neuronpedia 2026

Privacy & Terms Blog GitHub Slack Twitter Contact

Neuronpedia

APIAssistant AxisNEW Circuit TracerNEW Steer SAE Evals Exports Community Blog Privacy & Terms Contact

GPT2-SMALL · 8-TRES-DC · 659 ｜ Neuronpedia

Home
GPT2-Small
Transcoders Residuals
8-TRES-DC
659

INDEX

Explanations

the word "we're" in various contexts

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

case

-0.77

ppa

-0.77

mine

-0.76

anto

-0.72

rule

-0.71

ousse

-0.70

onso

-0.67

istry

-0.66

rench

-0.66

-0.66

POSITIVE LOGITS

 unable

0.88

 ecstatic

0.79

 glad

0.78

 unaware

0.76

 skipping

0.73

 attempting

0.73

 proud

0.72

 thankful

0.71

 concentrating

0.69

 conducting

0.68

Activations Density 0.025%

No Known Activations