INDEX

Explanations

The neuron fires on numeric literals (multi‐digit numbers) in the text.

comma-separated numbers

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

 kompati

-1.50

 VOOR

-1.48

 Euró

-1.45

 deklar

-1.45

PÅ

-1.41

 konfer

-1.41

 téléph

-1.38

TÉ

-1.36

 tille

-1.36

 intensif

-1.35

POSITIVE LOGITS

そして

1.25

1.20

 sogar

1.20

1.19

These

1.18

 incessant

1.17

 three

1.16

三种

1.14

1.13

 countless

1.13

Activations Density 0.070%