INDEX

Explanations

state or action

The neuron fires strongly on open‐class (content) words—nouns, verbs, adjectives—while largely ignoring 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

-1.79

of

-1.72

ⓚ

-1.35

-1.33

 admon

-1.31

 통해

-1.30

 uphe

-1.30

 esche

-1.27

 primeiro

-1.26

EEN

-1.25

POSITIVE LOGITS

 demostrado

1.45

 kupa

1.37

 this

1.36

ⓒ

1.34

༑

1.30

羕

1.29

нологі

1.27

this

1.27

any

1.27

 indican

1.27

Activations Density 0.167%