INDEX

Explanations

ending punctuation

The neuron fires on ordinary content words—common nouns, adjectives, and verbs—while ignoring proper names and pure function words.

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

 rito

-1.25

 réjou

-1.18

 mulia

-1.13

Estadio

-1.10

 améli

-1.10

 ilan

-1.10

駅の

-1.07

 teka

-1.05

 amargo

-1.04

 ventes

-1.02

POSITIVE LOGITS



1.24

">[

1.21

 Black

1.19

Glej

1.14

rootView

1.13

 prieš

1.11

 Nachbar

1.09

(){

1.06

})$,

1.05

KEY

1.05

Activations Density 0.327%