INDEX

Explanations

names of people

The neuron activates strongly on tokens that are parts of people’s names—especially capitalized first names, initials, and surnames.

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

our

-1.71

 their

-1.63

 your

-1.48

 this

-1.45

(^^)

-1.39

so

-1.36

 soyez

-1.35

 bicicleta

-1.30

!!!!

-1.29

!!!!!

-1.28

POSITIVE LOGITS

 ameste

1.58

<bos>

1.45

 lorsqu

1.38

桥

1.34

を用

1.21

 myriad

1.20

 opulent

1.19

ského

1.19

 contextLoads

1.18

mathrm

1.16

Activations Density 0.040%