INDEX

Explanations

each unit or group

The neuron primarily detects numeric tokens representing quantities or measurements 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

 then

-1.24

 using

-1.22

 अलावा

-1.15

 this

-1.14

but

-1.12

了很多

-1.11

so

-1.06

 manteiga

-1.03

 without

-1.02

 Johnson

-1.02

POSITIVE LOGITS

 fiecare

1.38

 búsquedas

1.37

 each

1.27

 moze

1.20

 sebou

1.20

都需要

1.17

 setiap

1.16

 smak

1.15

 habitaciones

1.13

 jednej

1.13

Activations Density 0.141%