INDEX

Explanations

editing, readonly, Liverpool, Seeking, We, Effect

The neuron fires on code tokens that refer to mutability or write‐permission settings—e.g. “editable,” “ReadOnly,” and “editing.”

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

for

-1.04

its

-0.99

双方

-0.97

all

-0.97

 semua

-0.96

 domésticos

-0.96

of

-0.95

เที่ยว

-0.95

 FÜR

-0.94

évé

-0.93

POSITIVE LOGITS

 olmad

1.15

 raffredd

1.14

 véhic

1.11

Hanya

1.02

mtable

1.02

 ohnehin

1.01

 себя

1.00

 réjou

0.96

Langkah

0.95

 дочь

0.94

Activations Density 0.000%