INDEX

Explanations

numbers and citations

This neuron fires on tokens that contain numbers (i.e. any alphanumeric or symbol string with a digit).

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

 that

-1.70

 nowego

-1.53

嫱

-1.52

 some

-1.49

 voegen

-1.48

 fewer

-1.47

gilet

-1.45

 those

-1.44

rifle

-1.44

to

-1.43

POSITIVE LOGITS

 даже

1.79

1.55

1.54

 всех

1.46

1.43

There

1.41

 حتی

1.38

".

1.35

for

1.34

Activations Density 0.239%