INDEX

Explanations

consent or dismiss criminal fine

The neuron is looking for numeric tokens (e.g. numbers, dates, case or section numbers, and other digit sequences).

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

 prét

-1.36

ῴ

-1.34

 ljud

-1.32

Focusing

-1.32

 ujedno

-1.31

 roś

-1.29

჻

-1.28

ὁ

-1.27

 talvez

-1.27

 frequentemente

-1.27

POSITIVE LOGITS

ּוֹ

1.56

on

1.54

וּ

1.52

 السياس

1.39

ַּ

1.34

 because

1.30

ִּ

1.30

וּ

1.30

with

1.27

 before

1.25

Activations Density 0.019%