INDEX

Explanations

court appeals

The neuron fires on named‐entity tokens—proper nouns and formal titles (e.g. vehicle names like “truck,” sports teams like “football team,” courts like “Supreme Court of …,” and other named institutions).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits



-0.72

juje

-0.72

 translocation

-0.71

 dinners

-0.71

HAUS

-0.69

건

-0.68

 transl

-0.68

็ต

-0.68

hausen

-0.67

heta

-0.67

POSITIVE LOGITS

 الحم

0.77

 apoder

0.73

hci

0.73

Sched

0.68

星の

0.68

idou

0.66

unehmen

0.66

atsu

0.65

hoenix

0.65

顷

0.65

Activations Density 0.106%