INDEX

Explanations

if you can't

The neuron fires on imperative or directive verbs (e.g. “try,” “use,” “please,” “contact,” “imagine”) that introduce instructions or suggestions.

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

 arrivent

-1.25

îtres

-1.23

 Meksiko

-1.20

 suivent

-1.20

 devraient

-1.20

おはよう

-1.18

 azonban

-1.17

めちゃくちゃ

-1.17

igné

-1.16

 donnent

-1.15

POSITIVE LOGITS

to

2.03

 other

1.80

or

1.45

it

1.27

1.20

其他

1.20

 during

1.20

 (“

1.20

 their

1.19

 used

1.17

Activations Density 0.095%