INDEX

Explanations

action or contribution

The neuron responds strongly to content‐heavy abstract nouns—especially words naming actions, qualities, or contributions (e.g. “work,” “activities,” “contribution,” “success,” “position,” “ministry,” etc.).

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

%">

-1.46

 menyebutkan

-1.43

ないです

-1.23

maila

-1.20

ρους

-1.20

хая

-1.19

𐍄

-1.17

什么呢

-1.15

 menegaskan

-1.12

tainer

-1.10

POSITIVE LOGITS

of

5.22

ของ

1.93

by

1.92

 của

1.80

 from

1.55

our

1.32

 terbaru

1.29

 people

1.28

and

1.21

 של

1.19

Activations Density 0.061%