INDEX

Explanations

. W

The neuron predominantly activates on standalone numeric tokens (digits or multi‐digit numbers), i.e. it’s detecting numbers such as page, section, date, or resolution identifiers.

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

برت

-1.10

९

-1.07

이나

-1.04

 Bakter

-1.03

や

-1.03

 aktu

-1.02

ता

-1.02

 napi

-1.00

 informacje

-0.99

rsiniz

-0.98

POSITIVE LOGITS

 Также

1.11

 الحكومة

1.10

 daarbij

1.04

 XXXII

1.02

 prič

1.00

同時に

1.00

 XXIV

0.98

 XXVIII

0.98

夯

0.98

OrWhiteSpace

0.97

Activations Density 0.016%