INDEX

Explanations

native of

The neuron fires on numeric tokens—integers, decimals, IDs, dates, page numbers, DOIs, etc.—essentially spotting any number‐like string 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

檮

-2.03

横に

-1.65

 endforeach

-1.59

 teile

-1.54

 haga

-1.54

 ajud

-1.51

甗

-1.47

-1.45

ैसे

-1.44



-1.42

POSITIVE LOGITS

of

2.27

1.70

In

1.66

 име

1.58

dė

1.54

、

1.50

odeon

1.50

 Aufenthal

1.49

 vorigen

1.49

斯

1.48

Activations Density 0.001%