INDEX

Explanations

suggests ability or role

The neuron fires on abstract, conceptual relation words—especially nominalizations and role/ability markers (e.g. “role,” “ability,” “importance,” “presence,” “being”).

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

 juſ

-1.25

 censi

-1.16

 raiſ

-1.15

',\

-1.13

がありません

-1.11

')

-1.11

 concerto

-1.09

 They

-1.09

-1.08

月下

-1.08

POSITIVE LOGITS

or

1.41

 Sugges

1.37

 only

1.33

 more

1.32

1.28

すべて

1.23

 лише

1.20

aruh

1.20

 Tells

1.19

 types

1.16

Activations Density 0.147%