INDEX

Explanations

made ofpurpose of

This neuron is detecting standalone numeric tokens (e.g. years, counts, measurements).

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

 covid

-1.39

 ไม่

-1.39

 récol

-1.36

Doesn

-1.34

 มี

-1.32

 alluring

-1.25

Biden

-1.25

extremely

-1.24

 empêcher

-1.20

Katalog

-1.20

POSITIVE LOGITS

in

1.82

at

1.40

 with

1.34

to

1.33

1.32

had

1.31

 began

1.30

 because

1.30

 despite

1.26

for

1.22

Activations Density 0.193%