INDEX

Explanations

numbers, Roman numerals, or letters followed by punctuation

The neuron primarily detects sequences of digits (numeric tokens) 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

is

-1.41

-1.39

ur

-1.28

en

-1.27

-1.22

-1.20

コチラ

-1.19

te

-1.17

POSITIVE LOGITS

 norvég

1.40

 copos

1.36

ってこと

1.31

 manquer

1.26

 paille

1.22

mû

1.21

 durer

1.20

 occuper

1.18

しましたが

1.17

 mús

1.16

Activations Density 0.002%