INDEX

Explanations

mathematical expressions

The neuron fires on long numeric literals (multi‐digit number tokens).

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

//{

-1.56

 kalt

-1.52



-1.48

˃

-1.46

ctober

-1.45

toDouble

-1.37

Ѕ

-1.37

☎

-1.36

♀

-1.36

 kanton

-1.36

POSITIVE LOGITS

and

1.91

mathrm

1.57

1.45

1.34

1.31

brtc

1.29

 einerseits

1.28

 durumda

1.27

 dezelfde

1.25

ときの

1.24

Activations Density 0.039%