INDEX

Explanations

parameters, descriptions, and types

The neuron fires on document‐formatting tokens—especially standalone numbers and punctuation markers (e.g. bullets “*” and code‐fence backticks).

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

 kulinar

-1.75

diğim

-1.37

emphas

-1.34

 doloroso

-1.34

vän

-1.33

毬

-1.30

kapital

-1.28

 chêne

-1.27

ığım

-1.27

 выбрать

-1.27

POSITIVE LOGITS

for

1.67

to

1.62

if

1.55

 there

1.38

衞

1.34

 required

1.31

 nutrisi

1.30

メダル

1.30

 бого

1.28

 acestei

1.24

Activations Density 0.026%