INDEX

Explanations

data warehouses

The neuron selectively fires on numeric tokens and list‐item markers (numbers, decimals, and their surrounding punctuation) in the text.

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

ּוֹ

-2.08

ַּ

-1.84

ְּ

-1.72

ֶּ

-1.66

ִּ

-1.63

וּ

-1.59

 épaules

-1.56

וּ

-1.55

 المسؤول

-1.45

᾿

-1.37

POSITIVE LOGITS

you

2.30

if

1.66

can

1.65

You

1.64

ׇ

1.63

 можно

1.63

 your

1.59

 ideas

1.50

 olabilir

1.44

 such

1.43

Activations Density 0.021%