INDEX

Explanations

The neuron fires on open‐class “content” words—nouns, verbs, adjectives, adverbs—rather than on common function words or punctuation.

consequence or condition

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

 seemingly

-1.59

or

-1.56

 might

-1.51

if

-1.44

 possibly

-1.39

 perhaps

-1.38

 There

-1.37

 relatively

-1.37

 sizeable

-1.36

 after

-1.34

POSITIVE LOGITS

iiiii

1.41

 dando

1.32

他们

1.26

 créer

1.25

iiii

1.25

 famoso

1.23

unfortunately

1.20

 dezelve

1.20

 hugely

1.20

primarily

1.19

Activations Density 0.100%