INDEX

Explanations

publish

The neuron strongly activates on occurrences of the word “publish” (i.e. references to publishing actions).

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

 have

-1.63

…」

-1.52

 bilden

-1.47

));

-1.45

艄

-1.34

jetos

-1.34

.”

-1.33

imli

-1.33

 adage

-1.32

霎

-1.32

POSITIVE LOGITS

’，

1.59

 زیب

1.46

硖

1.41

此时

1.40

سة

1.40

 riguarda

1.34

 richiede

1.34

かなりの

1.34

↵

1.33

has

1.29

Activations Density 0.041%