INDEX

Explanations

number year illustrations

the neuron strongly activates on numeric tokens (digits and numbers such as page numbers or years) 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

 sätta

-0.87

โรง

-0.84

韓國

-0.83

Tham

-0.80

韩国

-0.80

⾜

-0.79

 NUKAT

-0.79

スポーツ

-0.79

 പ്ര

-0.79

Ша

-0.77

POSITIVE LOGITS

 illustrations

1.45

 illustration

1.41

 Illustration

1.33

Illustrations

1.30

 opposite

1.13

 Illustrations

1.11

illustration

1.09

illustrations

1.09

 illustrated

1.08

Illustration

1.08

Activations Density 0.007%