INDEX

Explanations

numbers, statistics, and measurements

This neuron activates on numeric tokens—numbers or measurements (e.g. “19.5,” “213,” “1724”)—marking quantitative data 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

it

-2.27

 have

-1.87

</u>

-1.86

 will

-1.61

</th>

-1.50

 know

-1.48

 that

-1.47

how

-1.44

 azok

-1.41

 they

-1.40

POSITIVE LOGITS

几个人

1.49

着一个

1.41

<h2>

1.37

两天

1.37

羕

1.29

我真的

1.28

幼児

1.26

我这

1.26

still

1.25

 الأوروب

1.23

Activations Density 0.105%